CASE STUDIES: 'THEORY AND ECONOMICS - CASE STUDIES: 'THEORY AND PRACTICE IN AGRICULTURAL ECONOMICS A pafu#prescmted to the 41st Conference of the Au.vlralkm AgriculltJrf!. and

Download CASE STUDIES: 'THEORY AND ECONOMICS - CASE STUDIES: 'THEORY AND PRACTICE IN AGRICULTURAL ECONOMICS A pafu#prescmted to the 41st Conference of the Au.vlralkm AgriculltJrf!. and

Post on 08-Feb-2018




0 download


CASE STUDIES: 'THEORY AND PRACTICE IN AGRICULTURAL ECONOMICS A pafu#prescmted to the 41st Conference of the Au.vlralkm AgriculltJrf!. and Resource A-lm,~-getmmt Economics Society, Gold Coast, 23-25 .kmuary 1997 Jim Crosthwaite', Neil MacLeodl and BUI Maicolrn1 1. Depa11ment ofAgrkulfure attd Resource Managetnent, University of Melbourne, ParkviH~ Vic 3053 2. CStRU 1'rupicQi Agriculturet St Lu!!ia Qld 4067. Abstract Relevant research solves problems, and solving problems ill present day agriculture and natural resource management increasingly itwolves drawing on krtowledg from a range of disciplines. The mix of disciplinary knowledge appropriate to answes questions depends on the nature of the problem at hand. Research resource CnStraint means that there are trade-.f1ffs between the number of relevant cases which can. be included in an analysis and the di.sciplinary breadth artd depth brought to bear on each case. Thus there is a continuum of research methods from traditional agricultural economics dealing in a shallow way with large numbers of cases and drawing on. a narrow range ofdisciplinary knowledge1 focusshtg on. a few features of ea(fh case, to the classical business management approach which typically deals with few cases and draws on a wide rattge of disciplinary knowledg~ to analyse complex systems in depth. The latter approach, commonly called the case stQdy method1 has a useful role in natural resource economics research. Attention to . thet"a.ucal questions concerning design of the analysis can enhance considerably the value .of the output of case s.r~Jdies~ Introduction Case study methods are frequc~tly employed across a. wide range of social science disciplines including, for example, farm management; business management, marketing and psychology. Within these fields, they have long been regarded as both a legitimate and powerful way of explorirtg research and policy questions. For many reasons, not least th. imperatives and rewards of specialisation, agricultural economists have . generally favoured eonometric techniques de(\ling with a narrow range of information covering a large number of ca~s relating to. the issue in question. the hope is to derive some general conclusions abQQt some aspect of a large number of cases, itvestigated in a narrowly disciplinary and relatively shallow way. By comp!ltison, the case study rnetltod involves exploring fewer exampl~$ at greater depth. The aim of explanatory case studies, as distinQt from e)(pJoratoJY and. descriptive ones, is to investigate a small number of operating systems front many angles and in depth! to obtain insights into the likely impacts of (:banges to different but similar systems. We would argue that once accepted as a legitimate research approach, c.ase studies can be employed by agricultural economists to make a valuable contribution to research and policy dcveloptnent. The discounting of the case study method has genemlly taken two forms. On the one hand, case study research has commonly been seen as an ~easy' option that may be usei-ul for teaching purposes, but otherwise inferior to surveys because of limited ability to provide substantive insights beyond the particular case. That is, there is a widely held prejudice against the apparent ability to generalh;e beyond single or, at bestt from a limited array of cases. On the other hand, the case studies that have been carried out with appropriate disciplinary breadth artd depth are generally regarded as being complicated, messy aud~ all Utings considered, simply too hard to carry out and 1neaningfulty use. However, while there is a fair element of truth underpinning the latter viewpoint, it does not fully explain why the technique is largely ignored in agricultural economics yet widely used in other disciplines. In this paper, we outline the nature of case studies, their strengths in comparison to other teclmiques, and provide some examples of the type of research for which they are .most appropriate. The ability to generalise from appropriately designed. case study r!!search is also discussed, in particular the fundamental difference in generalisation from case studies {analytical generalisation) as opposed to that from. statistically~b;L..c:ed techniques (statistical generalisation). However, in order to successfully generalise from case studies, the true nature and rofe of case study methods needs to be better appreciated and their conduct pursued with. no less. diligence and rigour than other empirical research techniques. In particular, the design phase irt case study research is critical to their successful conduct, and is a. topic of central concern to this paper. It .is illustrated by reference to one of two case study projects that ate currently beit1g developed by the authors which are incorporated within the socio-economic compcment of the National :Remnant Vegetation Pl~ogram. being jointly funded by the Land and: Water Resoutces Research and Development Corporation and the Envirortm~llt Australia Biodiversity Group. The exarnple case study project is centred on livestoek grazing man~gement, and its effects on production and conservation perfonnance in the grassland$ and native pastures of southern New South. Wales and northern Victoria. the second project (which is rtot discussed in this paper) has a similar focus but is located itt the sub-trorical woodlands of southern sub-coastal Queensland. The project c~ example is used to 'illustrate the various phases in the case study design process, from theoretical specification, of tbe issues to dev~loping tests for validity and reliability. Issues in choosing a method for our projects TtJt policy problem ,\ The extensive grasslands and woodlands of Austtalia have majot degradation ptobleiJl~ which ate ~f concern from both a production and a conservation viewpoint (e.g~ totbill arid Gillies 1992, .. Department. of Environment, .. Sport and Territories 1996). The pro(>lems vary . frolll region to region. While from, an economi~t' s v;,ewpoint, there rtlJY be in$ufticient pn=cision about the extent and lotation ofthe problems, the scientific tQttuntlllity and ,policymers are in general agteen1ent of a need to address this problem with some urgency,. To tbis .,nd, some major programs a.-e now in place in an attempt to address the problems {e.s. Land and Water Resources R&D Corporation Remnant Native Veget~tion Program, Environ.ment Australia Biodiversity Group ''Save The Hush" Program) . Front a general production viewpoint, the major problems involve loss of vegetative cover (especially from perennial grass species); soU otgrutic content and physical structure decline, and other elements contdbuting to nutrie,nt and water cycling~ as well as intrusive problems such as salinity" acidification. and erosion (e.g. Mcintyre and Mcivor 1996). From a conservation viewpoint~ there are rel$tively .few gl'a.sslands and woodlands that can still be characterised as natural.eCC)systems, and those that remain have a conservation signific(lttce well beyond their size. The~ remaining areas are subject to .many influences (e.g. dearing, weed invasion, overgrazing) wbieh will lead to their loss or to irreversible degradation. Grassy woodland ecosystems ~e under":represented in fonual. reserve systems (e.g. national parks, conservation areas), and this situation is unlikely to be redressed (financially, politically) within the .foreseeable future. Morver, there is a genuine doubt concerning the effectiveness of attempting to preserve such representative ecosystems within a formal reserve system anyway (e~g. Mclntyre 1994). The policy probleru arises becau.~e bio-physiciil, social and ecoJtomie aspects are inter-related (e.g. Iiardngton, Wllson and Young l984). The consequences are tnanifest in biophysical terms but also in effects on tanning practices and farm viability. The. causes are primp.riJy socio-econom.ic in character butj once set in train. the changes in bio-physical proeesse$ t@ke on a life of their own. The soluti.ons: will depend on scientific research; but the .possible outcomes,. how they are to be achieved and the pace of achieving them greatly influenced by socio-economic factors and. the action. of many indi.Vidual laud .resource managers. To really understand the likely impact of new tecbuologies, policy .initiatives and/or ex.ternal develo.pments (e.g. climatic change; rnarket changes; trade developments) on patterns oi :resource use that may impact on conservation values, the decision-making processes of individuals and the rich context within which this occur$ ,needs > be better understood. One issue or harrier to real progress~ however, remains the belief (rationally grounded. or otherwise) that community con8rvation objectives are necessarily in cnfli.ct with .Production objectives of' both individual land managers, .jf .not the community itself (e.g. Department of Environmentt Sport and territories .1996)~ De$pite the obvious resource management and :policy implicat.ions; the underlying ._reasons for this petcelvd conflict remains largely unresearched. Som irnportllnt conaideration for '"curce mnas.ernent mrch. In addressing land resource man~gernent problems. issues of scale are important, from the vieWpOint of both ecology and economics (e.g. HatringtQn, Wilson and Young 1984,. Brown and MacLeod 1996). fanner ~lecisions about either exploiting or conserving teOlll#Dt vegetation are typically made at the whole farm enterprise. .level. HQwever; most *'&ricultutal R&D (including agricultural economics a$sessment of tril data) is conducted, at Slnaller scales (e.g. plots1 land classes and occasionally paddock~) and, thereby~ f~Us to ~ the context within which such decisim)s are typi~ly made. Extension.oftbe result$ c>fluchRAD are, . non-surprisingly, . typically pi\dlt,;U at the same inappropriate ~e$ . e~ technology transfer failure problems (MacLec)(t and. Tylor l99S)+ To.~fully ~. problems in land use decision-making, an Qbvious starting point i$ UJ Jet the ~e riaht We would argue that this Jmplies a detailed understanding bf the whole property re.~o~ structure; managment (technologicat) systents and the SOtio-economic onttxt of the managers (e.g. age~ dependants, interest, affCJrdability, beliefs). Addressing these issues are, it~ tum, believed to hold the key to itnptoving adoption of sustainable grazing management (e.g. MacLeod and Taylor 1995). A r(!search metbod which supports the concept ()f exploring both underlying processes and cm,tcxt is cleatly required (e.g. Pettigrew t 985). Conservation management research will ideally seek to combine cross .. secth:mal (what's hnppea\ing now across a range ofcases) artd J~1ngitud.inal (is it stable over time) elertumts. This is because the coo text fot conservation on Jarms is unlikely to be uniform or static over time or space. For example, a farmer may base a given decision on present levels .of wcalU1, current. prices, policies or understanding .of the government of the day . As these variables can readily change\ so docs the d~ision context and, therefore, the likely decisions made and their consequent outcomes. Studie~ which effectively address these changes are important. The .key questiot1s. conccmins . management and the context within which the technial parameters of tnanagcment ate laid down must be included in these, Fc:;r example, preferences, attitudes, social motes, opportunity valtJcs for fatm labour and natural or tnanufa:ctuted resource endowments that underpin much decision-making. These ate commonly ignored or asstm1cd away in technical efficiency studies, wbich increashtgly underpin a !e-em.erging interest .in benchmarking and best ptactic in l~nd r~source management (e.g. Clark and. Filet: 1994). 111e management of retntlant mttiv0 vegetation witt targely depend on fanrt-family goal$. Fanners have fairly complex. choice (unctions. (e.g. Cox and Ridge 1997, Dunn et al. 1996), which ate not ope.n to simple study or assumptions of rationality (e.g. profit maximisation). Identifying policy SThe nature of~ case studl Whit.,. ca .. atud'"? Case: studi~s and the caS study method have. trtlditionaUy been bard to explofC because of.the limited thcoretic;~l. and t1pplied treatments they b~we received in the past. Generally, the SJ)Ccialiscd textbooks on experimental .methods and design have either ignored the method ot confused it with a topic or field to which they have been applied (e~g .. ethnography). A key f~iling h~ been to critically dcfitle the technical features .of ~ study ~1tategies that speciiJcally distinguish them fronl other research strategies (e.g .. survey$, histories, eleperiments). Were these technical features better understood, then the role aod power of carefully conducted case stt~dy research would be better appreciated and, hopefully, utilised . 'rhis deficiency ha~ partially been redressed* in recent times, through (amongst {)thers) Jhe effort$ of Robert Yin~ a leading scholar in. the domain of case study research theory and practice in the social science.s (e.~. Yhl l98la. l98lbt 1983~ 1989. 1993). Yin (1989) has argued that much of the ~'bad pressH enjoyed by case study methods typically stems from poor definiti-on ofcase studies as a re~rearch strategy. Through his WJiting, t~aching and .. research, he has sought to provide a clear definition of case study methods, clarify their role and appropriateness within the potential. arra~ or empirical stmtegies for addressing research ques,tions, and promote rig our and discipline into their conduct. Yin(l98 ta) provides a technical definition of a case study as an empirical. inquiry that: {a) in.vestigates $. ctmtcmprury ph~nomenon within its realllfe C(Jnte~t; when (b) the boundar;es between. the phenomenon and context are not clearly evident; and in which (c) multiple sources of evidence are used. The italics are our own emphasis. Managen1ent of remnant vegetation is a. C()ntempQtar.V phenomenon within a. fanning context and is, therefore, subject to many and vario"'~ influences (e.g. price levels. availability ot'' feed on, other parts of the fann. available family .labour etc.). Tbe boundary between management of a tract of remnant vegetation (phenomenon) and the whole farm. (context) is often. difficult to distinguish. For' e~plei while an examination of typical farm rttords may ,provide 11scful information on stocking intensities .and a. limited array of practices o11 different .. pam. of a. property; they are rarely collected at a resoll1tion suffici~nt .to distinguish revenues and costs attributable to ~h part of me enterprise on which tationat as$essments on reSQurte allaation between ~nservation and produc.tion rnight be based. Moteov~t, the majority of landholders may not pi~: much signifi\$ance on remnant or native vegetation, .simply seeins it as part. of~ to .. l bundle of resources with whi~b to make land use deci$ioats. Finally, the combi01don of phenomenon and context: are uniq~ on each farm atld, more, widely, fc>r each re$0UI'Ce '* decision problem ~nrronting ~soure managers and policy n1akers. Other resateh appmacbes typicaliy .handle C(lQfetnpotalleQU. data ....... ~-cog~t relationshiP' with Umited efficacy or cffi.:iency. For eurnplc, ciMSic ex~ ~ is to divorce .Phenomena from context through. "controlled" environmental conditions. While ttt~ditional survey rncthods atguabl) might be en1ployed to e~plore both pknomenon llld context; they typically St.-ek to lituit the number of varhbfcs canvassed. This, .however, limits their insights into conCext, which lllay be critical to the research question being explot:ed. An hnporumt distinction needs t( be made betwf!en case study mc:thods and, the: general domain of qualih:ttive research. J>art of tbe perception of ci!SC study n1ethods being usoft" lies in the tnistakcn categorisation ofthc method within the mote geneml domain of qtutlitative ~ opposed to quantitative research. Without wishin~ to denigrate the use of qualitative resc:~b methodst which can gellcrate p,lwedul insights into resource use decision-making pt~esses (e.g., Pettigrew 1985~ .Patton 1990),, the hquaHtati"equllntitative,delineation has little bearinQ ()rt cboice of emJlirlcal research approach (Yin 1989). Ca..'ie studies can be conducted entirety on the basis of quantitative data. coll\!ction and analytical techniques. Therefore)\ while ~ studies frequently rely \.>O the collection and analysis uf qualitative data, this .is not a necessary chatateristic of the method. Beyond lheir research role. case studies arc n vehicle that can be readily used for the dissemination of useful extension n1atcriltl based on ugricultural economics research. In seeking to est;tblish sustainable grazing management systems, managers. appear more readily convinced by demonstrations of practical success implcrnented at the whole property scale (Mucteod and 'rayJor 1993, 1995). Choosing betwHn c atudi and othtr tHhrdqu Yin ( 1989}, while recognising ~onsiderable overlap between the characteristics of various enlpirical methods, suggests that a choice between them might more ~tionally be made against three conditions; vlz: tu) the type ofreseatch question being posedz (b) the extent of conttt1l a researcher has over actual behaviaural events; and (c} the degree offus on conlemporar)t as opposed t() historical events. The first condition really boils down to the simple '~whot what, where, when,. why and how" questions ()n which rese~ch is typically .focused. While any of these questions can be handled by most te~arcb approaches, this is aec;;omplisbed with varying degrees of' efficicmcy* Fof example, '~who", 4'what" and "Wberen questions might be best addreSSd tbmugb $Ui'VC'YS or historical accoQil.ts. The more interesting (from our perspective) 'how'~ and "why" que$tiQJ1S, which are explanatory rather than exploratory or descriptiv~, are well addressed by taSC study .methods. However, other methods such as fonnaJ experiment., and historical ~unts also often employed to address this type ofre$eai'Cb question and 5Cl Yin. (1919) sussests the second and third Qnditions provide the n:essary discrimination. liHstQrical accounts are best u.-sed where there is no $COpe for control ovet or insiaht into contemporary events. Experiments, to be useful, require an ability to control and ..,..Upulate events in a. direct, ~ise: and systenuatic fashion which rarely can be ICCQinpJiahed beyond laboratory conditions. Yin (1 989,. 1993)'. therefore, ideattifios appro;riite niche fot . .-o .6 study methods in. re~ch $ituatiotts whi~h deal with contentpotaJY events in: "~doh behaviour of the people or ~yst.erns at 1he centre of the research problern cannot &e manipulated. This rote is also suppor1ed by two investigative t~bniqUC$ (sources ()f evident>.e) that arc of limited use to other methuds "' dirt~ct observation and systematic interviewing .. These tc~hniques cart be usefully applied to other SQUrces of evidonce (e.g. documents, .a.rt~hival materials~~ surveys etc) to provide the multiple sources of evidence that are the third technical charaf.:teristic of case study methods (previous sub-.section). Multiple sources of evidence~ and the eatlt study method, lll'f! important in researching fanning systems because they are senerally acknowledged to be comp1eKt and their perfbnrttu1ce is influenced b)' many purposive and atl hoc m~nagement decisions .. This Qecurs within a context of many Hl .. det1ned or poorly understood feed-back loops and considerable uncertainty. Controlling such systems with an aim of defining the contribution of various factors (e.g. native grassland) is very difficult~ especially in the presence of limited information about: the functioning of the plant .. animal interface. Oynamic ,processes and chnnge are also characteristics of fam1ing systems~ and the case study method can .capture the key elements of' these prC(lrttinuing to see case studies within the context of ~s3tflples" is a mistake and .rt. root' cause for failure in the useful application of the method. Individual cases are :not sampling units in any strict statistical sense The correct context for gcneralisirag beyond immediate case findings is that of theory development and generalisation. to theocy. SimHar to clas$ic scientific e>."}ledm.ents (whose generalisation ls rarely queried), valid case study design will ideally be based on a well-grounded theory and set of :propositions to 'be tested by the .calse. The findii1gs are then generalised to that theoretical base according to the ensuing d('gree of support the .empirical . findings ptovide to the testable . propositions. Tb u of t~stable theories and rival theories support$ this pr~ss; which Yin (1'}89) likens to level ,one i'lfercnce. which is the underlying ba~;is of most laboratory and field e~perhnent-; in the agricultuntl. sciences. lf the empiri~al fi.ndings support the theury . or better suf)p(>rt a rival theory. progress is made in tbeoey development. Cast study methods ..-e particularly important in expanding knowledge of theoreti~l propc$ltions and h~these$ in: those situations where {a) the oontext is important and .(b) events ~t be manipulated as in a classic experhnent: (Yin. 1993). It follows that confidence would ordinarily increase IS the cmpiri~. findinss *" abo found to apply to multiple cases, consistent with th theoretical context from which the: m.t. ~. was drawn (i.e. analytical. generalisation) .. In the same vein, the use tJf multiple. cQQ studies (like their direct analogue of multiple ex~riments) can enhance lhe analytical generalisation pr~ess through repUcation, especially when the empirical. findings of the lldditiout dl$CS suppOrt a given theory while contradicting a well justified rival thC()ry~ The usc ofmu1tiple cases should not be confused with increasing Hrepresentativeness" of "s&ntples" which ..,.,lies to statistical generali~tion logic {lttvel IWl inference .. sample to poptdation) .. already argued to be inappropriate to case study methods, Why the case study appro1ch Wit chosen For our work with valuation and rnanagement ofremnant vegetatiotl, the case study approach was selected because of the critical role played by issues of botla process and context that underlie the central questions being addressed. That is) questions of how and 'Why and against what background or within what environment decisions are made concerning sus~inable management of native pastures within an. extensively graz.ed gra...;sJand ol' grassy woodla:nd context. li.!nce both projects aim to undcrslahd the farm system - what the technology is .md how it fits {in the existing farm layout, the farming systern, intonmanagement'' ~ aQe, interest, affordabiHty, be!Hefs). The insights of this ecological and economic R&D should produce {Xllicy recommcndati.ons to promote the sustainable management of these pastures fot conservation and production .. Neithet a sample survey or a tnodeUittg exercise can adequately captute the es~nce of management, the monetary and non-monetary goals, the !amity income needs, the cbatactcd!!ti'!~ r,f each land type on the farrn etc. A case study can do this! .albeit with limited information for tn~ny aspects. The case study can also utilise a. range of sources and cross .. check results to ensure validity.ln order to obtain a. picture of feed availability, the grassland: economics ptoject will draw heavily on farmer records and memory for stock rnov.,ment between paddocks over 12 months. Other sources used to double-check this will include research trial data, observations of district farmers and site inspections by agro-tomists. The rich insights ofcase studies; which involve extensive dialogue with decision-makers. is argued to allow for specifying and disseminating potential solutions that are feasible both operationally &nd economically. Phases of case study rrch .. the importance of sound d .. lan Once a problem or issue has been :;elected for research, all research is anducted in phases that typically. but not always, follow a general seqnce of establishing a research desisn, collecting the daU., analysing it, and re.porting the findings, ease study :re.,ch is no differcntt These phases are now co.nsidercd~ R ... arch&atign Research design is the vehicle throqgh which the data collected in any study or e)(periment and the conclusions ~ nec:essarily linked (Yin .. 1989) .. All empidqal rese.tch i$ driven by so111e fonn of rese~ch design Wbich is. either .. expliobly $late0 or, at .J~ implicit front observation of the actual tttnduct of the rescarh . ln .~ ]atter OQCt ~ de;ip are poorly thought through (or even ignored), the quality of the research and aenetal validity .of tM ensuing conclusions is open to challenge, Case study research is no ('7(CCption 41Pd' seems ~ be particularly vulnerable to problems of poor or. Hmited formal design. Many re~bers selecting a ~\~ase study approach" seem to commence data collection witbQUt adeqUA~ly s~cifying their proposed rnethod or giving serious consideration W tb~ desien "' po~$ibly under the belief that case studies ate an e8ploratory toot that precede$ some more formal eX(X!rirnent or survey ifanything "'interesting'' emerges (Yin 1989). This is unfortunate logic, as w~ are here arguing that case study hlthods are \'alid research loots in their own. right mtd are not merely pre-experiment e)(pJotations (although they c(ln be), 1'hereforet tQ be successfidly applied! the critical design phase, must be well-conducted and adequat~ly sllcifl~d (Yin 1993). The key to es~bUshing a good research design for case study .. based research (or any research methodology for tbat matter) is to follow a logical process \1f linkin~ data to objectives, conclu~ions to data and~ thereby, linking objectives to conclusions; In following such n proccsst Yin ( 1989) identities five. components or elements of u case study research desigu that are particularly important. These are as follows: (I) J>rescnting a clear and adequate specification of the theoretical issues and, from this, the proposition..'i that underpin the study. (2) Clearly de11ning the unit(s) of' analysis, including possible sub .. units if these ate ~ f. .,,,attanted ,. (l) DecJding on the appropriate nurnber of cases to explore within the study .. .. ~. f4) Clearly speci.fying the selection criteria for choosing the case studies. ~. ,, l> . (5} Choosing an appropriate and effective data collection and aualyt;is str~teSY +-~ . . ,l .,.. (6) Oevel.oping appropriate tests to ensure the validity atJd reliability of the approach taken in conducting the ca~ study. These elements are brietly outlined, and then the specific approach taken in the example, native grasslatld case study is illustrated; Specifying the theoretical issues necessadly begins with a. cleat outline of the specitic is$uea of concern fot the propQsed study. Related to this process of considedns issues to lddtes$ is the basic need to clarify the underlyins re:;earch questions .. ie. the ~'who, where, what, why, and hown question$ addressed in a previous sub--section. It . .,.,. already been suasested INf case. study methods ideally address uhow'' and '~why't questions rathet than the other ijtpe~ and S(l this is .~.simple t~chec;k on th~ bMic ;sppropriatet'le$& uf adopting. the ~ study JPPfOfieh, From this exposition oftbe issue$ ~d ~sutance of the ~fficacy ofthe apptoaf;b, ptOpoJiUons can be developed which ,;;"plain the situation. Rival proposi;in$ may also be velope4 lt. this stage to be e"plored within a study, The format propo$ltiont~ .;an then be te~ vi data. coll~ction and t~nal)!Si$ ofthe results .. R~scarcbers ~ommon}y fail to clearly define and/or stick to a un'it of analysis thttt is appto.priate to exploring the theoretical issue . ~r proposition. that; fonns the base . of the research study (Yin 1989). A clear definition of the UJ1it of analysis is neces~ to place titnt bmmdaries on th~ subsequent study. to develop televant nnd precise hYPOtheses, and to guide the collection of data. l1oor ot intprecisc definition ~fthe unit of analysis will typically lead to results that. lack rigonr and/cH at best ate descdpdve rather tbaJl explanatory (Yin 199:1). An associated problem o~curs where data are collccfed tbr sub .. units (e.g .. a paddock) and the eusuin~ analysis is conducted exclusively at. that level~ rather than being drawn together at the or.iginal unit level (e.g. the fanu) for which the issue to be researched wus originally identifi~d (Yin 1989). th~ unit of analysis need. not t~late to some specific physical entity such as an. individual, group or institution .. Nevertheless, some physical ot personal basis is often preferred as the unit boundaries are often taidy clear. Ahernative units of analysis might include a specific policy (e~g. production quotas) or how it was implelllCilted (e.g, national or state basis)t an event (e,g .. severe drought or market shock) at er;:t (e.g. pre versus post .. deregulation of the finance set~tor)t The key t.o determining the appropriate unit of analysis remains the re$Carch proposi.tions that have been. defined for the study (Yin 1989). Number of c:s Unlike statistical stunpHng methods there is no h~d and fast rule concetning the minimum number of cases to be selected for a given research. project. This results front the inherent difference (noted before) between the logic ofanalytic and statistic4tl generalisation. Sek:ction of the number of cases within a cas study is necessarily influenced by the ~purpoSes of the study, the reseatch propositions that ate to be tested and the J~vel of confidence 'that is required in the findings., the confidence bh~g Nf~rrcd to hete is not that represented in n statistical sense (e.g. 9S%t p(c) construct a tich theoteti~al tram~work Oil the basi$ of the individual c~ses whicb .can then be generalised to a broader theory ()t onde.,tanding of tb~ pbenomen~ of b1terest to the study (e.g. the process and context issue.,;), embedded case studies, which may b applied to either single or multiple Ca$e designs, sinlply refer to instances wher.e the theotetical propositions are best expbned via multi--tiered units of analysis. Fur example1 a study of small and Jarge fanns may still. wish to further explore the implications of age or family size by selectin~ sub-units from both fann types according to these demogtJtphic factors. An important issue that is related to tbe selection of multiple c~es in a case study design is that of replication. In arguing that case study methods appropriately generaJiae to theory rather than tt.1 population;;~ Yin ( 1989) makes an important distinction. between literal and llu.wretica/ replication. These two forms of replication are standard features of classic~ CXJ~dmentntion. Litcr~l replication involves the selection of particular cases {experiments) on the ba.~is that they should predict similar t(!SUlts. Theoretical re.plication, on the other hand, irwotves the selection of cases that might produce contrary results .. but for reasons that are ~onsistent with an underlying theoretical proposition. In thi~ way, multiple cases which incorporate both literal and theoretical replicates can be used to extend advance tbeor.r~ This is a fundamental cm1sideration for using the method and. ultimately, successfully linking the data and conclusions to the ptopositions. In some ca')eSs Amdysis of data. will be; carried out htdepend~ntly tbr each c~~ study, i)()th l'~latins bk lP the ol.liectives ttnd drawing out policy bnpUQntiOtlS ''~ each additionAl cuse i.s otnpl~ted~ the results nre clu~cked to sec if they replicute the fin\lings in th~ previous cases, Once all cttse~ are completed, cross .. case eonclusioa)s can be drawn (Yin lV69)~ For tmy particular research design ... case study or (.)thcrwise ... there are essentially four basic f.ests of logic thnt might be applfed to nssess its quality. These are: (a) construct validity .. upproprinte definitions and operaUoni.ll measures for the theoretical propositions being studied; {b) internal validity .. appropriateness fot' estnblishing crecJible causal relationships; (c) external validity .. convincingly spccif.ylng the domain to which the findings can be generalised; und (d) reliabilit,v .. ability to repeat the findings if the same m~thods \:tc are upplted .. Yin {1989; 1993) hn:t proposed the tblt(lWing means to incorporate these within a case study design: CmtstfliCI 'validity. Using several ways to measure the key variables (constructs) in th~ study is an importnnt way to overcome possible pr(lblcms of jnaccuracy. Multiple sources of evidence nre cl~ar:Jy needed when little infotmntio' is available nn some aspects of native, pasture or farm manaacmel1t. Intentttl valid/iy. 'the theory must be internally consistent. This requires Qarefut specification of the units of analy!iis so that th~ study does not slip from one unit to, nnother1 and use of ri.vnl theories which are tested ugainst the collected data. Ex/ental vcc,'idily. This requires 11apecilication of thcoreticLtl relationships; from which. gemmdisations can then be made" and iR a mnjor justificadoo for using .multiple case study designs, particularly those that are cu1beddedt Relhtbility. Fottnal protocols are n~cessary to ensure that procedures are. consistent actm the native grnsslnuds project. Case example ... The native grasslands project Background to the project The project is cc:mtrcd rn specialist grazing propetties ht southern New South Wales and northem Victoria which have either some remnant native grasslands (largely on. the riverine plaif1s) or some native f)asture (largely on the fonnerly wooded slopes, h.itls and: tablelands). Remount grasslands are generally recognisP.d to hold significant c()nServation value because they contain both a diversity of plant species and supPQrt a range of threatened. species (e.g. Plains \Vandetcr). Native pastures are predominantly composed of native grasses and are of particular interest fbr their possible role itt preventing and Qverconting deg . u:lation of agricultural land through salinity, acidification, erosi.on mtd soil structure decline on the poorer classes ofland. The two year project cornnlCJlced in August 1996. It fotlows an earlier exploratory project in which i:'Ul11ers on 28 properties in south-eastern Australia were interviewed and some preliminary budgeting on managf.mtent options for native grassland was undertaken. The objectives of the project relate to ( l) clarifYing the potential economic role ofnative grassiand, (2) identifying other factors that ate important to landholders ht managing native grassland, and (3) looking at appropriate. policy instruments (focussing on incentives) for achieving conservation gouls. Three other objectives (complementary to the t1.rst objectiv~) are to (4) produceregi.on..:specific economic info.rrnation; {5) develop dc.cision~makhtg methods for uncertainty; and ( 6} develop a decision analysis package. A final objective is to (7) assist: in adoption oftbe study findings. While all objectives will be addressed within the one research desigt1, the first objective relating to quantifying the on-farm tole of native grassl~d is used to clarify the case study research design process. Theoreticallu and research propoalticU1s .h1 order to quantify the et,onomic role of native grasslands, comparisons need to be drawn between the economic petfortnance of uresent management sttategies and alternative strategies. Where farmers are managing for conservation at .Present, tbe project needs to dctel1tline whether this management: can conthlUe without significat\t opportunity cost. Conversely, where conservation management is not occurring, the opportunity costs of changing .managem.ent to accommodate conservation needs need to be estimated. The profitability and (mancial feasi.bility of different manage1nent strutEl8ies needs to be assessed on a whole farm business basis, rather than for an investment in isolation. 11lis levl of analysis is most meaningful to fanners, and it can. account fh~ the inter-relationships between native grassland and c>tber parts of the fann. It also give$ \ibc: range of production advantages and weaknesses that have been attributed t() native grasslpnd a significance that would not apply if they were examined indh,idualfy, ta Efte~ts of management changes on ecosystem functioning anil the land and water resoutces are difficult to determine, and prior research into these is Hmited. It is possible to test alternative strategies as ~bypothcticals' with parameters detennined through consultation with case property managers, region&l stakeholders and technical specialists (e.g. conservation biology, pasture ecology$ gr~ing systems). Two rival propositions are to be used to guide the research. HO: If most income is not derived from the native grasslands, fanns with native grassland can meet farrnfamily inconte needs while nianaging native grassland within conservation constraints .. provided wool prices are not very low. Hl: lfmost income J! derived from the native grasslands, fanns with native grassland cannot meet farnt~family income needs while managing native grassland within conservation constraints .. unless wool prices are high. Several research questions can be derived from these statements. They are: What. disposable surplus (net farm income, operating profit; cash flow) can farms with native grassland generate? How can it be explained that some &1nus with native grasslands can meet income needs, while others cannot? How can income needs be met if current levels are insufficient? Taking account of climatic and price uncertainty, can it be done by; (a) changing management or use. ofgrassland areas? (b) other means (on other areas of the fann, or off~farm}? Whether income needs can be met by managing within conservation constraints, both for short~term retention of species richness, and for long~teml system stability which rnay be required as the underpinning of the production system. In order to clearly distinguish farms with native grassland accotdbtg to ability to .meet fal111 family income needs, the proportion of income derived from native grassland will be. set at two levels: below 50% and above 70%. Wool prices over $6.50/kg are regarded as .high. Very low wool prices are regarded as under $4.00/kg, The main unit of analysis will be tb~ farm business. This is bro-d tnougb to cn(}otnpass decision choices aoout investment and work off .. fann. Sub-units will include: the farm, the land type (e.g. introduced pasture, native grassland), and the paddock (the basis fot estimating stocking on each land type). Data will be collected, for each of these su~units and. some nnalysis will be conducted at these levels, :However~ the ctiticai task is to explain the effects of dit!erent manage111ent strategies for native grassland at the farm. business level. Numberofcaaea The ptopositlons are to be tested for variation in. several factors. Some do not require represetttative selection of cases to be tt.!sted, others do. These are physical conditicns and proporti.on of fann . ..:famHy income derived frotn native vegetatit;m areas .. These are .explained as variable l aod 2 below, lmd the implications for the number of case studies are .illustrated. The effect of fatm size and fann managcmertt1 variables 3 and 4, can be tested without fanns on these criteria. Variable 1 .. lund typl. Topographyt soil type and rainfidl are largely independent of human action and affect carrying. capacity and management options. Case study fanns will be selected for two land types .. (a) riverine plains and (b) slopes, hills and tablelands. Similar t'esults (literal replication "' Yin 1989) can be expected from. case studies t;clccted in each. land type, hence a minimum. of two is needed to test thls. Testing .for differences (theoretical replication"' Yin 1989) between the two land types will require a minimum nf one case study in each. ln fact, the project has been designed so cuses are selected for two sub-regions (ensuring repteseutatio.n of both Victoria atld New South Wales)within each of the two land types,.. a total of eight Gases. Variable 2 proportion of fann.tamily income derived from native vegetation areas. Across sub-regions. results will be compared for case studies with similar proportions of income (low or high) derived from native vegetation areas (literal.replication). :Resuit.!:i from c~~e studies with low and high proportions of income derived from native vegetation aras will be compared to see 1f the differences are explained: by the hyp4.1lbeses (theoreHcal replication). Half the case!; wiU be selected for a low proportion. and .half for high. V ar.iable 3 .. farm size. Farm size for any gi.ven physical conditions directly influences income~ Farm size will not be directly tested through selection .or replicates based on fann size. Instead, other data linking farm 'financial resl1lts to fatm size will be used in conjunction with case study .resul~c; to estimate the effect of farm size on capaci.ty tt~ manage; native>. grassland within conservation constraints. ' Variable 4 grassland areas arc managed within the whole .farm; system. M.anagernent of native grasslands. and their use in conjunction other parts of the fa,rm. will vary greatly from fam1 to fann and is likely to have an important effect on farm income. Tbe etlect .of management will be estimated by identifying alternative management options for .each farm, und testing the economic and finUncial implications of eac~ S.tction c,...ria Speci.fic criteria to be used in selection of case fatnts which. are important for this objective include: farms rnust have native aw~land; varying proportions offaml business income derived from the gative ~sland areas; availability of information about stocking the native g.-assland paddocks compared to others; the fanns should at least one land type eg introduced pasture as. wen as native grassland. or have .native grassland across several. different soil types; native grassland with high conservation values should be present on some fanns; the landholder should be interested, abk to give access to records and availabl'e for interview;, the farm should preferably, though not esserttially, be linked to a native grassland research project (e.g. Community Grasses) or part of a LandCare group~ research information should preferably~ though not essentially, be available for similar grassland sites to those on the case farm. Data c;ollectlon strategy The prQ.ject will involve in-depth semi-structured interviews with landholders and possibly other family members as well as mt1re formal collection {e.g, structured surveys, inventories etc) ofextensive technical information relating to the property (eg financial records, stock and paddock records etc.). Data .requirements are Hlustrated in the Attachment. ln order to obtain a of feed. availability. the grassland economics project will draw heavily on farmer records and memory for stock movement between paddocks over 1~ months. Other sources used to double-check this will include .research trial data,. observations of district fanners and site inspections by agronomists. Data about management and stocking of paddocks will be collected frotn. at least two of the case study fanns for the previous year, and will tben be collected on a continuing basis for at least another year after the initial collection. This approach will give at least two full years of data. As management of grasslands, particularly in drier regions, is greatly influenced by ~nal CQnditi(Jns, a longitudinal' dimension to the project is important. The following steps summarise the approach. Detennine each land type on the case fann, and allocation of paddock$ to land type$; Estimate c~nt stocking rate in terms of livestack months (bun) for each land t)rp=; Estirnat~ monthly carrying capacity of each Jarnt type :for different levels of conserv.ion constraint and for different seasonal conditions; Decide feasible management alternatives, at1d monthly stocking implications; Generate economic and. financial results for eah management alternadve by inc()rporating production, cost and income into a spreadsheet; Test for sensitivity of key variables; Analyse the results for the each case fanu ing th1, criteria indicated above to determine whether the hypotheses have been confirmert for that case and to draw polity implications; Compare results to previous cases; Draw cross .. case conclusions. The following criteria is applied to interpret the findings. If the cash surplus (adjusted for yearly fluctuation) available to the family is within 90% of the required income, it will. be judged "adequaten. However, iflt :is less than 70% of the required income~ it wiU be judged to be ''inadequate'\ Any farnt investment must meet '~nonnat' economic and financial criteria, if it is tp be regarded as superior to current management. The following measures wiU be used: economic {net present value, inte.mal tate of return) and financial (cash flow, breakev~n period, peak debt). Tnting the dign for validity and rtliabUity, Con~vtruct validity Where pOSsible tnultiple $0urces of evidence wHl be used. Fanner opinion on stock held in each paddock and stock rnovetnents \\ill be matched to numbers in each. mob. Movements of mobs will be correlated with significant events such as shearing. Finally, such information. Will be correlated with farm records where available~ Costs and income will be veritle!d against other data sources such as gross margin handbooks compiled. by agriculture departments. Potentiat stocking rate for different levels of conservation constraint Will be. estimated on the basis of multiple sources .. botanist: opinion, agronomist opinion, fanner opinion; available research data. Each case report, . and . spreadsheet, will be presented so that the chain of evidence 'caD be followed frpm initial data to final. conclusion .. Each. case reJ)Ort will be presented to the case farm owner and to other key informants for review. lntern41 and externalvalidity Careful comparison of the replicales will detennine whether lite ... l teptitateJ do in f~M;t ~show like results, and whether the th~rctically different rep1icat~ show diff\CeS that, :_, consisteot with th~ underlying theReliability ReHabilit)r ~nd consi$tency of the study will be addressed by documenting k~y aspect~ of the: study and by maintaining a data base for each case study wbicb is independent of the case report. Discussion/conclusion The paper originated in the need to find a suitable ntethod for &Uliilysing on .. fartn land mnnagement .issues. The case study method bas many advantages over otb~r techniques. The percei"ed weakne.sses of case .studies have been .addressed and shown not to be. substi.lntive, though case study research is by no. means easy. Tile errors that are commonly made in their application have been. outlined. rhese problems can be overcome by using a good research design, which includes a stage of theory development and application. The step.s in research design were illustrated by application to the natural rewurces problem that lead to consideration of the case study apprShulman, and R. 'Price, eds.) .Land' and Water Resoutce$ Re$Catcll &. Development. Corporation, Canberra (ln press). Harrington, O.N., Wilson, A.D. and Young, M.D. (1984). Management of Au.,tralia's Rangelands. CSIRO~ Melbourne. Land nnd Water Resources R&D Corporation (1995). Socio-cctmomic aspects ofmaiutaining natiw! w:gettllimt an agri'cultmtll lund .. Occasional Paper No 07/95. tWRRDC, Canberra. MacLeod~ N.D. (l997). Effective strategies for increasing the suitability and adoption of complex technologies for sustainable grazing land management. In: Improving Management of Sustailrable R&.D Technology Tran.ifer, Vol. 2. (A.D. Shulman~ and R. Price, cds.) Land and Water Resources Research &. Developmcrtt. Corporation, Canberra, {ln press). f\1acLeodt N.o. ( l997). Case studies .. a useful tool for integrating socio-economic data into systems research. Contributed puper UJ sth International Grassland\ (im:grcss. Jllintll!JJCg .. Sa~tkaiOon, s ... I 9 .July (J n press). MacLe:od. N.D. and Shulman, A.D. (1994). Best practice for technology transfer - Avoiding lost opportunities through improved integration .of R.D&E, Contrih~~ted paper to 8th AilS I .Rrmgeland Soc. Confonmcc Katherine. June .2 1""23 1 pp 31:4-4. MacLeod, N.D. and Shulman .. A-~J. (1996). What is wrong with conventional tecbnolQgy transfer practice for Au~tralian rangeland R&D? Contributed paper to 9th ilu.vt. !lfJngelmtd Soc~ Co fifo renee .f(,rl Augusta. 44 .. 2 7 September pp 189-90. MacLeod, N.D. arid. Taylor, J.A. ( 1993)~ Sustainable grazing practices in Queen$1and, Arl.~trallanFarm Manager 3(6):6 .. 9. MacLeod, N.O~ ru1d Taylor J.A. {199S)~ Perceptions of beef cattle prodUtS and scientists relating to sustainable land use issues and their impUcadons for technology transfer. RPngelantiJoun'lal16(2):238 .. s3~ MacLeod, N.D., Shulman, A;D. and Taylor, J.A. 0995). Participatory R;O&E projects ot co!laborative alliances: Resolvin~ complex problems with multiple users. Preedings of 5th lntf!rnationa.l Rangelands Congrc~s, Salt uik~ City 23-28 July 1995. Society 'For Range Management: , Denver, Co pp 3323. Mcintyre, S. (1994). Integrating ~grlculturalland ... use and rnanQcment for conse.vation of' a native grassland flora in a varlesated Jand~ape. Pac:Jfic Colr$ervation 1Jioi1Jgy l :236-44. Mclntyre, S .. and Mclvor1 J.C. (l996). /)IVI!f'N/ty and sustainability In rangelaflilllvestock production. sy$temt. C'I'C8 .Project B ..'litJn paper for till /AJntJ. and Wat~r Rl.!.tources R&D Cor/H)ration. CSIRO Division. of Tropi~ crops .mel PIStUt'CI, St Lucia. ., Pampel, F .. and Van Es, J.C. (1977). Environn1ental quality a11d issues of adoption. research. Rural St ~ii, "' ... '( ._.,....., . _..., >-" "'\I~- ~'Tr~ """''""'" =~--. income by source. debt financial viability Profitabibty Extent tn which income net.~ ail! Ult1 f~ fann-.hmily .famtt=!j:Onls. intervie\\'5 11'itn Interviews ~ith hmity fantl1!r(sJ .Profitability bi enterprise- Byinfercnce~farming production, purchases. sales approach.. fannittggoals &. sto:::l data;. -oveib(ads ~t ofJeUance ~nillc fann Income needs. (age,lifecycle-.. lifestyle_ oblip+Joos tdebt. dependentS, dividend payments)) Undty~ :film r..s.aps :;fai:mer kno'14'1C&'f&c. Wp~Cc~by botanistt~ Effect of~ manacemcntoo .flmi business ~ .toatribut".on .type; benefitS: oflllbvt' s.pks ~ }lcaJntsR JlrMCr ~ Soiltype.spcci:s. ~ By addihono~in& (il tbaracter".stii:.s. COl'ISCn'ltion status~ po&tnlilllstuckiD$. UtiJisatiort over year 000009620000096300000964000009650000096600000967000009680000096900000970000009710000097200000973000009740000097500000976000009770000097800000979000009800000098100000982


View more >